Re-running installer to add a second model does not update `OLLAMA_MODEL_PREF` in `.env`

[Raze3478]

When the installer is run a second time to add an additional model, the .env
update path checks only whether LLM_PROVIDER=ollama is already set:

if ($existing -match 'LLM_PROVIDER=ollama') {
    Write-Host " AnythingLLM already configured for Ollama." -ForegroundColor Green
}

If the condition is true, the file is left untouched. This means
OLLAMA_MODEL_PREF retains the value from the first install run and is never
updated to reflect the newly added model.

A user who installs model 5 (Llama 3.2 3B) first, then re-runs the installer
to add model 1 (NemoMix 12B), will find AnythingLLM still pointing to
llama3-local with no indication that the configuration was not updated.

Steps to reproduce

1. Run install.bat, select model 5 (Llama 3.2 3B)
2. Confirm anythingllm_data/storage/.env contains OLLAMA_MODEL_PREF=llama3-local
3. Re-run install.bat, select model 1 (NemoMix 12B)
4. Check .env — OLLAMA_MODEL_PREF still reads llama3-local

Expected behaviour
The installer should update OLLAMA_MODEL_PREF to the first model in the
current selection, or prompt the user to choose which installed model
AnythingLLM should default to.

Proposed fix
When .env already exists with LLM_PROVIDER=ollama, use a targeted
find-and-replace on the OLLAMA_MODEL_PREF line rather than skipping the
file entirely:

(Get-Content $envFilePath) -replace 'OLLAMA_MODEL_PREF=.*', "OLLAMA_MODEL_PREF=$firstModelLocal" |
    Set-Content $envFilePath -Encoding UTF8